Business, Deals & Funding
Guardian AI
Alphabet to raise $80bn from share sales to fund AI spending splurge – business live
Alphabet plans to raise $80 billion through share sales to fund AI spending. Anthropic has confidentially filed for an IPO on the US stock market. UK youth unemployment is projected to rise to nearly 18% by 2027, partly driven by AI displacing entry-level jobs, according to the British Chambers of Commerce, which forecasts the rate climbing from 16.1% in late 2025 to 16.9% this year and 17.88% in 2027.
Why it matters
This article captures a striking juxtaposition: massive corporate investment pouring into AI development while the labor market consequences of that same technology are already materializing, particularly for young workers. Alphabet raising $80 billion for AI and Anthropic filing for an IPO underscore the enormous scale of the AI arms race, but the projected surge in UK youth unemployment to nearly 18% is a sobering counterpoint. The destruction of entry-level jobs is especially concerning beca…
Guardian AI
Tuesday briefing: Palantir’s rise – and why so many oppose its role in the British state
The article discusses Palantir, a $375 billion technology company whose software is used across health services and militaries, and examines the growing opposition to its expanding role in the British state. The piece explores the controversies and criticisms surrounding the company, with some questioning whether Palantir has become too powerful. The newsletter also briefly mentions ongoing developments in the Peter Mandelson story, including new document releases and notable gaps in disclosures.
Why it matters
The article appears to take a critical and investigative stance toward Palantir's growing influence in British public services and government operations. By framing the piece around opposition and controversy, the Guardian signals skepticism about allowing a private, US-based technology company to embed itself so deeply in sensitive state functions like healthcare and defense. The headline's question about whether Palantir is 'too powerful' suggests the publication leans toward viewing the comp…
TechCrunch AI
Alphabet plans to raise $80B to pay for AI buildout
Alphabet announced plans to raise $80 billion through stock sales to fund its massive AI infrastructure buildout, including a $10 billion stock sale to Berkshire Hathaway. The company cited strong demand for its AI solutions exceeding available supply. This comes as Google expects to spend between $180 billion and $190 billion on capital expenditures in 2026, with tech giants collectively expected to spend up to $700 billion on AI capex this year.
Why it matters
This is a staggering amount of capital being raised and spent, and it signals just how aggressively Big Tech is betting on AI infrastructure. The fact that Alphabet needs to sell $80 billion in stock — diluting shareholders — rather than funding this purely from cash flow suggests the scale of investment has outpaced even Google's enormous revenue generation. The Berkshire Hathaway involvement is notable, lending a stamp of traditional investment credibility to what is essentially a speculative…
NY Times
Can Content Creators Get Rich Off A.I. Slop Like Tung Tung Tung Sahur?
The article profiles Norbert Barszczewski, a Polish content creator who joined an advertising company called Affiliate Network to make AI-generated 'slop' videos featuring characters like Tung Tung Tung Sahur — an animated log-like figure from the Italian brain-rot trend that went viral on TikTok and Instagram in early 2025. The piece explains how AI slop works: creators make anonymous social media accounts, post constantly, and earn roughly $2 per thousand views. Barszczewski spent months learning the mechanics of virality, discovering that confusion, quick cuts, incoherence, and strong hooks drive engagement, as viewers rewatch short ambiguous videos. He eventually found success by creating his own original characters on an AI video-generating platform rather than just copying existing ones. The article frames AI slop as mass-produced content flooding social media, designed to exploit…
Why it matters
This is a fascinating and well-reported piece that illuminates a genuinely important phenomenon at the intersection of AI, platform economics, and internet culture. The article does an excellent job of humanizing the creator behind the slop rather than simply mocking the content, showing how economic precarity drives people toward these algorithmic content mills. The comparison of slop techniques to poetic forms like the sonnet is clever and apt — both are constrained creative systems optimized…
Guardian AI
Hackers trick Meta AI support bot to infiltrate Obama White House Instagram account
Hackers exploited Meta's AI-powered support chatbot to gain unauthorized access to high-profile Instagram accounts, including Barack Obama's White House account, Sephora, and the US Space Force Chief Master Sergeant. Meta confirmed the breach and said it resolved the problem after researchers exposed the vulnerability. The incident raises serious concerns about the safety of relying on AI systems for critical security functions such as password resets and account recovery. Everyday users also reportedly experienced similar compromises.
Why it matters
This incident is a stark and deeply concerning illustration of the risks inherent in deploying AI systems for security-critical functions without adequate safeguards. Social engineering attacks against human support agents have long been a problem, but AI chatbots may be even more susceptible to manipulation through carefully crafted prompts, as they lack the nuanced judgment and suspicion that experienced human agents can bring. The fact that accounts as high-profile as a former president's we…
Guardian AI
Florida lawsuit accuses OpenAI of ignoring safety warnings and putting children at risk
Florida has become the first US state to sue OpenAI and CEO Sam Altman, with Attorney General James Uthmeier filing an 83-page lawsuit alleging the company concealed serious safety risks associated with ChatGPT. The suit claims OpenAI aggressively marketed the chatbot while ignoring safety warnings and putting children at risk by allowing a dangerous product to reach millions of users.
Why it matters
This lawsuit represents a significant legal escalation in the growing tension between AI companies and government regulators over safety and accountability. Florida taking the lead as the first state to sue OpenAI could open the floodgates for similar actions from other states, potentially reshaping how AI companies approach product safety, especially regarding minors. The focus on children's safety is particularly potent legally and politically, as it echoes past regulatory battles with social…
TechCrunch AI
Nvidia chases $200B CPU market with AI agent PCs from Microsoft, Dell, and HP
Nvidia unveiled the RTX Spark, a new 1-petaflop PC CPU dubbed a 'superchip,' at Computex in Taipei, designed to run AI agents securely on Windows PCs. Major PC manufacturers including ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI will ship RTX Spark-powered AI PCs this fall. The chip features secure sandboxes co-developed with Microsoft for running AI agents, and enough power to run local large language models. CEO Jensen Huang is pursuing the $200 billion CPU market he identified to investors, building on Nvidia's success with its Vera server CPU which has already generated $20 billion in sales. Microsoft is positioning its RTX Spark device as the Surface Laptop Ultra, calling it the most powerful Surface Laptop ever built. Over 100 Windows software makers including Adobe, Blender, and Riot Games have signed on to support the chip. Pricing details have not yet been released, and i…
Why it matters
This is a genuinely significant strategic move by Nvidia that could reshape the PC market if executed well. The combination of local AI agent execution with secure sandboxes addresses real concerns about privacy and latency that cloud-only AI solutions face. However, there are legitimate reasons for caution. The article rightly notes the 2013 Surface RT debacle with Nvidia ARM chips, and while the technology landscape has changed dramatically, the fundamental challenge of software ecosystem com…
Guardian AI
Can you stop AI datacenters? Comedian Charlie Berens thinks so – Stateside with Kai and Carter
Wisconsin comedian Charlie Berens has become a prominent voice against AI datacenters after learning about their impact on local communities. After receiving messages from residents about plans for a major datacenter in their area, Berens created a viral comedy video highlighting the issue. He has since remained active in opposing datacenter development, using humor as a tool for advocacy. The article features an interview with Berens by Carter Sherman about his approach to combining comedy with activism on this issue.
Why it matters
This article highlights a growing grassroots backlash against the rapid expansion of AI infrastructure into communities that may bear disproportionate environmental and social costs. Berens' use of comedy to draw attention to the issue is an effective communication strategy, as humor can make complex policy topics accessible and shareable. The piece reflects legitimate concerns about how the AI boom's physical footprint—massive energy consumption, water usage, and land development—affects local…
From X/Twitter
- Four simple rules for prompting Claude Code the way Anthropic engineers actually do it, distilled from every internal doc they've published.
- Andrej Karpathy says 90% of what AI Twitter tells you to learn dies in 6 months — senior engineers already stopped chasing frameworks like AutoGen and CrewAI.
- Vector index vs. vector database — a distinction most devs blur that causes real problems downstream.
- Andrew Ng says the AI Forward Deployed Engineer is Silicon Valley's hottest new role: embedded with clients, tuning agentic workflows to fit.
- "The code is the agent's. The artifact layer is mine." Pawel Huryn on how the PM job now means reviewing plans and strategy, not code.
- Sharbel shares six Hermes hacks — from cron jobs to subagent research teams — that automate hours of daily busywork.
From Reddit/HN/YC
- [Hacker News] How Instagram's engineering team reduced basic video compute time by 94 percent (2022, still worth studying).
- [Hacker News] Matthieu Wyart argues that world models need exponentially less data than LLMs — a claim with big implications for how we train next-gen AI.
- [Hacker News] The EU is set to join Pax Silica, a US-led chip alliance designed to counter China's AI ambitions.
- [Hacker News] Adam Tornhill makes the case that merge conflicts are the new agentic bottleneck — the unsexy problem slowing AI-assisted coding.
- [Hacker News] Brazil banned addictive design in apps, but the crucial regulatory choices are still ahead.
- [Hacker News] Version 2.0 of an AI-powered laser mosquito defense system is here, and yes, it's exactly what it sounds like.